sugg: AI research centers

[Zuikaku]

Please consider to reduce number of research centers the AI builds.
Although it is nice to see them build these, this needs to be toned down a bit. I suggest them to be able to build only 70-80% of research centers they are capable to build now.

I also suggest that research centers can not be captured and are destroyed when location is captured. These were either destroyed by retreating forces or were not being able to work properly during occupation for various reasons (defected or lost scientists, destroyed or lost equipment and/or documentation). That woul'd also solve issue of too much reseach centers after one region occupy other.

Please, consider to do this for the next update if possible.

[Balthagor]

The first suggestion is fairly easy. The second, not as much.
Task 23714

[Balthagor]

How many research centres do you think the AI should be trying to build? I think that once a region is over 800M population, it should be aiming to have 12-15 research centres. Around 150M+ 6-9 centres is a good number.

[Zuikaku]

How many research centres do you think the AI should be trying to build? I think that once a region is over 800M population, it should be aiming to have 12-15 research centres. Around 150M+ 6-9 centres is a good number.

In my game US have 11 in 1930 (1914 campaign). Think that is a bit too much (if we want to stay in some historical scale of research of the period). In any case 9-10 most. Are research centers dependant on GDP?

[Balthagor]

IIRC GDP/c affects rate of investment. If they do less than that, a human player could easily outpace them.

[Zuikaku]

IIRC GDP/c affects rate of investment. If they do less than that, a human player could easily outpace them.

But human player (without house rules set) can easily otpace AIs anyway.
But having US (and soon UK) AI completed nuclear bomb test by 1930 seems just wrong to me.

[Rosalis]

I have seen Russia in GC2020 with 50+ research centers in 3 years just by taking over half of europe, or South Africa with 35 research centers in 4 years just by taking over a cpl of African countries. If you dont think that's broken, the whole tech stuff is broken. In the end they just go bankrupt after causing huge inflation. 10 slots is all you need. If you cant let the AI destroy/scrap, then dont let them build. At minimal give player the option for ai to build or not.
I would actually like to see how this will effect economy compared to SR2020, and i think alot of players would prefer that, research centers show what is going on in economy of SRU i think. The option is already there, but i dont want to go thru 200+ regions just to change an option (and also block repairs i think), could this be implemented in starting options?

The cap between ai and player could be cloosed in many different ways, without hurting anything. Like you take over troop designs or techs from regions you capture or researching in the future gives a penalty and in the past a boost. It isnt that hard, right?

Locking the ammount of research centers to population is very unrealistic btw. Its often the smaller countries that come with innovations, and they already weak as it is. Countries like Norway, Canada, Denmark, Netherlands, Belgium, Slovakia, etc. should never be surpassed by African coutries just because they have more population. GDP/C, alliances, education, law enforcement, all should play a role i guess, if not in ammount of research centers then in research effectiveness.

[dax1]

the number of research centers should consider: population and level of education

[SGTscuba]

the number of research centers should consider: population and level of education

and available finances as efficiency costs go up with the number of centres, independent to the number of techs being researched at any one time.

[evildari]

the number of research centers should consider: population and level of education

and available finances as efficiency costs go up with the number of centres, independent to the number of techs being researched at any one time.

and especially weird as efficiency costs are usually extremely higher than the project costs (and iam talking about factors - not simple percentages).

On the original issue - well i like that with this, some ai run 1st tier countries manage to outtech my 3rd tier region.

[YaYo7]

Balthagor, In the new patch there will be changes in the construction of the research centers for ai? So that ai does not spam them unnecessarily

[Balthagor]

Not directly, but some work was done on the AI assessing how many of various facilities it needs. Might impact this indirectly.

This thread is low priority, there's no real consensus above about what our target should be. As I reread the thread, the concern seems to be more about centres "inherited by conquest" than what it builds.

[Rosalis]

I think there is concensus. Superpowers 20 max, powerfull nations 10-15 (UK, France, India), modern nations 5-10 (Netherlands, Australia, Canada). Big population/rich countries 5 (Egypt, South Africa, Saudi Arabia Turkey), rest 1-5. These targets would be for 2020 without gaining more territory. Ofcourse if Saudi Arabia annexes middle east and Turkey, well it would be a powerfull nation atleast. Could just be fixed with adjusting starting research centers, but people want ai building to RP 36/CW scenarios, so... there has to be a max or fixed with events.

In my lastest save game US got 48 research centers without capturing any territory. Result? They research everything! Even stuff with the biggest penalties, and still got 30 spots open. Seeing how much research costs impact economy, you must be a terrorist not seeing this as high priority.

Anyway its good to hear you guys are looking into AI scrapping. This feature was introduced with comment that it could easily be adjusted if it was too much, everyone is saying its too much, besides evildari, but yeah in his mods with his tech tree its very cool.

[YaYo7]

Example: US starts in CW1949 sandbox with 7 research centers, if over the next 50 years America builds 10 research centers. 17 centers this will be the optimal amount by ~2000 year. This is very close to the US value in the sandbox world2020 with 19 RC. I agree with Rosalis that 20 research centers are good value for superpowers.

[YaYo7]

How many research centres do you think the AI should be trying to build? I think that once a region is over 800M population, it should be aiming to have 12-15 research centres. Around 150M+ 6-9 centres is a good number.

How do you like this option?
300M-800M+ popuplation and gdp/c higher than world average = 20 research centers.
300M-800M+ popuplation and gdp/c lower than world average = 12-15 research centers.
100M-300M popuplation and gdp/c higher than world average = 12-14 research centers.
100M-300M popuplation and gdp/c lower than world average = 7-10 research centers.
20M-100M popuplation and gdp/c higher than world average = 8-10 research centers.
20M-100M popuplation and gdp/c lower than world average = 5-7 research centers.
0M-20M popuplation and gdp/c higher than world average = 4-7 research centers.
0M-20M popuplation and gdp/c lower than world average = 2-4 research centers.